Identification of Two Phylogenetically Related Organisms from Feces by PCR for Detection of Salmonella spp.

Abstract

Two previously reported PCR methods were evaluated to determine whether they are as sensitive and specific as conventional culture methods in detecting Salmonella spp. from feces. Bovine and equine feces were enriched overnight in brain heart infusion broth and assayed using PCR methods and primer sets described by other investigators. A total of 774 fecal specimens were tested using a primer set (invE-A primer set) that amplifies a region spanning the invasin E and A genes of Salmonella enterica serovar Typhimurium. A subset of these fecal specimens (306 of the 774 total) were tested using primers (hisJ primer set) that amplify a portion of the histidine transport J gene. The PCR required 24 h to obtain results, whereas it took 5 to 7 days to identify Salmonella spp. by culture. PCR detection of Salmonella spp. using the hisJ primers and the invE-A primers had a sensitivity of 93.3 and 80%, respectively, and a specificity of 85.6 and 98.6%, respectively, compared with bacterial culture. Amplification of 42 culture-negative fecal specimens (of 306 total specimens) generated a DNA fragment that corresponded to the molecular weight of the amplified hisJ gene. The hisJ-generated amplicons from six culture-negative and six culture-positive specimens were sequenced and analyzed using DNA sequence alignment and phylogenetic analysis software. A neighbor-joining dendrogram of the DNA sequences of both sets of hisJ amplicons revealed two distinct groups—one group of amplicons from culture-positive specimens identical to the hisJ gene of S. enterica serovar Typhimurium and a second group of amplicons from culture-negative specimens that were more closely related to hisJ of S. enterica serovar Typhimurium than to other hisJ sequences present in nucleotide databases.

Salmonellosis is associated with medium to severe morbidity and even mortality in cattle and horses and thus represents a major economic and productivity loss in the large-animal industries (5, 7, 12, 21, 23a, 32). Salmonella-infected animals in a veterinary hospital, dairy, or other animal populations must be quickly identified so that they can be isolated from other animals and spread of infection can be controlled (12, 13, 15, 20, 22, 29, 30). Outbreaks of Salmonella infections have occurred in large animals in several veterinary teaching hospitals and have resulted in significant expense and, in some cases, temporary closure of the hospital (15, 18, 23, 25, 29, 30). A more rapid method for detection of Salmonella spp. from fecal and environmental specimens could facilitate biosecurity procedures and minimize outbreaks in these environments.

Prompt identification of Salmonella spp. from feces is a challenge to laboratory diagnosticians due to the lengthy time required to culture and identify this bacterium from feces and environmental surfaces by conventional culture methods (8). At the Colorado State University (CSU) Diagnostic Laboratory, fecal specimens are enriched for Salmonella spp. by inoculating feces into enrichment broth, incubating the broth for 24 h, and culturing on plates of selective and differential media (31). Suspect Salmonella sp. colonies are selected based on morphological characteristics, and their identity is confirmed by substrate utilization. Environmental surfaces at the CSU Veterinary Teaching Hospital are monitored for Salmonella spp. by wiping a moistened, sterile sponge (HydraSponges; International Bioproducts, Redmond, Wash.) over a surface and then incubating the sponge in thioglycolate broth for 48 h. A sample of the bacterial suspension is removed and plated on selective and differential media, and suspect colonies are identified as described above. Because the average time between acquisition of a fecal or environmental surface sample and definitive identification of a Salmonella isolate is between 5 and 7 days, it is often difficult to preclude spread of infection in a hospital or herd situation when conventional culture methods are used.

Due to the need for identifying Salmonella spp. in a rapid and reliable manner, several laboratories have developed PCR-based tests for detecting salmonellae from feces (1, 9, 11, 19, 33). However, interpretation of the results of PCR tests can be ambiguous compared to those of conventional bacterial culture. For example, a previous report indicated that 17% of fecal specimens from healthy horses were positive for Salmonella spp. by PCR, but bacterial culture failed to recover the bacterium (10). Since bacterial culture has traditionally been the “gold standard” for identification of Salmonella spp. from fecal specimens and since feces contain PCR inhibitors (34), the goal of this study was to determine whether results from two previously reported PCR assays would agree with those from culture results of feces collected from hospitalized equine colic and food animal patients. Two different primer sets, invE-A and hisJ, that had been previously reported by other laboratories for the PCR identification of Salmonella spp. directly from enriched feces (11, 26) were chosen for testing. These PCR assays were reportedly successful at detecting Salmonella spp. from feces but had not been compared extensively with culture methods in a veterinary diagnostic setting (11, 26).

A total of 774 fecal specimens were cultured and assayed by PCR to detect Salmonella spp. in this study. All 774 specimens were tested using the invE-A primer set, and 306 of the 774 specimens were simultaneously tested using the hisJ primer set in the PCR. A total of 640 equine fecal specimens originated from animals hospitalized at the Veterinary Teaching Hospital due to colic. Although some of these equine patients had signs of clinical salmonellosis (fever, leukocytosis, diarrhea), the majority did not present these symptoms. Fecal specimens were obtained from bovine patients (n = 134) admitted to the Veterinary Teaching Hospital regardless of the presenting complaint, and these specimens were included in this study.

For PCR amplification, fecal specimens were enriched for Salmonella spp. by inoculation into brain heart infusion broth (Becton Dickinson Microbiology Systems, Sparks, Md.) as described previously (26). DNA was extracted from the enriched bacterial suspension (2.5 μl) using GeneReleaser (Bioventures, Murfreesboro, Tenn.) per the manufacturer's recommendations.

Amplification using the invE-A primer set consisted of adding the 10-μl volume of DNA extracted by GeneReleaser to 40 μl of a PCR mixture that contained the following (final concentrations are given): 1× buffer [16 mM (NH₄)₂SO₄, 67 mM Tris-HCl (pH 8.8 at 25°C), 0.01% Tween 20], 1.5 mM MgCl₂, 100 μM (each) deoxynucleoside triphosphate (Promega, Madison, Wis.), 1.0 μM (each) primer, 0.001% gelatin, and 2.5 U of Taq polymerase (Bioline, London, England). The PCR was achieved with 35 cycles of 94°C for 20 s, 56°C for 20 s, and 72°C for 20 s. This was followed by one final extension step at 72°C for 2 min, 30 s.

Amplification using the hisJ primers consisted of adding the extracted DNA to 40 μl of a PCR mixture that contained the following (final concentrations are given): 1× buffer [16 mM (NH₄)₂SO₄, 67 mM Tris-HCl (pH 8.8 at 25°C), 0.01% Tween 20], 1.5 mM MgCl₂, 200 μM (each) deoxynucleoside triphosphates (Promega), 1.0 μM (each) primer, 0.01% Triton X-100, bovine serum albumin (0.1 mg/ml; Gibco/BRL, Gaithersburg, Md.), and 2.5 U of Taq polymerase (Bioline). The PCR was achieved with 40 cycles of 94°C for 35 s, 66°C for 35 s, and 72°C for 45 s. This series was followed by one final extension step at 72°C for 10 min. DNA extracted from a Salmonella enterica serogroup C₁ serovar Infantis isolate from the CSU Veterinary Diagnostic Laboratory Salmonella spp. collection served as the positive control.

The PCR products were examined by electrophoresis through a 1.4% agarose gel that contained ethidium bromide (0.5 μg/ml). A 100-bp DNA ladder was used as a size reference for the PCR products (Gibco/BRL).

For culture of Salmonella spp., 1 g of feces was inoculated into tetrathionate brilliant green enrichment broth (7.5 ml; BBL tetrathionate broth base [Becton Dickinson and Company, Franklin Lakes, N.J.], 0.1% brilliant green dye) containing 100 μl of iodine solution (30% iodine, 25% potassium iodide [wt/vol] in water) and incubated at 42°C for 16 to 20 h. The enriched bacterial suspension was streaked onto XLT4 (PML, Wilsonville, Oreg.), brilliant green with sulfadiazine (Becton Dickinson and Company, Sparks, Md.), and Hektoen enteric agar (Becton Dickinson and Company, Cockeysville, Md.) and examined for colonies that morphologically resembled those of Salmonella spp. at 24 and 48 h after incubation at 33 to 35°C. Suspect colonies of Salmonella spp. were streaked onto tryptic soy agar containing 5% sheep blood, and their identity was confirmed with a MICRO-ID Microbiological Identification System (Remel, Lenexa, Kans.). All Salmonella isolates were serogrouped using BBL Salmonella O polyvalent antisera (group A-E/Vi; Becton Dickinson Microbiology Systems) and serotyped by the USDA National Veterinary Services Laboratory (Ames, Iowa).

A comparison of culture and PCR results using the invE-A and hisJ primer sets is shown in Table 1. Sensitivity and specificity values were calculated as indicated previously (24). Five fecal specimens were PCR positive with both primer sets but were negative for Salmonella spp. by culture, supporting the hypothesis that PCR could detect Salmonella spp. present at a low level or could detect nonculturable or nonviable Salmonella spp. from feces. Also, in one animal, bacterial culture of feces was negative for Salmonella spp. on the first day of hospitalization but was PCR positive using both the hisJ and invE-A primer sets. Five and six days later, both bacterial culture and PCR of feces from the same equine patient were positive for Salmonella spp., suggesting that the PCR may have detected Salmonella spp. present at a level that was undetectable by bacterial culture on the day of admission.

TABLE 1.

Comparison of bacterial culture and PCR for detection of Salmonella spp.^a

Primer set	Result by:		No. of specimens
	PCR	Culture
invE-Ad	Positive	Positive	36
	Negativeb	Positive	9
	Positivec	Negative	10
	Negative	Negative	719
hisJe	Positive	Positive	14
	Negativeb	Positive	1
	Positivec	Negative	42
	Negative	Negative	249

^ainvE-A and hisJ primer sets were used for detection of Salmonella spp. from 774 and 306 (of the 774 total) fecal samples, respectively.

^bConsidered a false-negative PCR result, as culture is the standard method.

^cConsidered a false-positive PCR result (see text) since bacterial culture is considered the gold standard.

^dSpecificity, 98.6%; sensitivity, 80.0%.

^eSpecificity, 85.6%; sensitivity, 93.3%.

The invE-A primer set amplified a 457-bp fragment, as expected, from 36 of 45 culture positive specimens (of 774 fecal specimens tested) (Fig. 1). PCR with the invE-A primers had a level of specificity that was almost equivalent to that obtained with bacterial culture; i.e., there were fewer false-positive PCR results with the invE-A primers compared with the hisJ primers. The 9 specimens (of 45) that were culture positive but PCR negative were considered false negatives, since amplification of purified genomic DNA from at least four of these isolates using a spaQ primer set resulted in the production of an amplicon in real-time PCR. The remaining five Salmonella isolates were nonviable after extended storage and could not be tested. Therefore, the lack of amplicons in these samples was not due to loss of pathogenicity island I, as has been reported previously for environmental isolates of S. enterica serovars Senftenberg and Litchfield (14). We attempted to enhance the sensitivity of the PCR with the invE-A primers, but it was not possible in our laboratory to increase the sensitivity above 80% with this primer set.

FIG. 1.

Agarose gels of amplicons generated by PCR amplification of enriched fecal specimens using the invE-A (457-bp amplicon) and hisJ (496-bp amplicon) primer sets. (A) Results using the invE-A primers. Lanes 1 and 6 contain molecular weight standards (100-bp DNA ladder). DNA used for amplification was as follows: lane 2, DNA extracted from culture-positive clinical specimen; lane 3, DNA extracted from culture-negative clinical specimen; lane 4, negative control (distilled water); lane 5, positive control (purified DNA from S. enterica serovar Infantis isolate 978-05817). (B) Results using hisJ primers. Lanes 1 and 9 contain molecular weight standards (100-bp DNA ladder). DNA used for amplification was as follows: lane 2, DNA extracted from culture-positive clinical specimen; lanes 3 to 6, DNA extracted from culture-negative clinical specimens; lane 7, negative control (distilled water); lane 8, positive control (purified DNA from S. enterica serovar Infantis isolate 978-05817). Arrows indicate amplicons produced in each reaction.

The hisJ primer set amplified a 496-bp DNA fragment from 14 (of 15 total) culture-positive specimens and from 42 specimens that were culture negative (Fig. 1). Fifteen of the 306 specimens tested using the hisJ primers were positive by bacterial culture (Table 1). In the single fecal specimen (of 306 specimens tested) that was culture positive but PCR negative, lack of PCR amplification was due to inhibitors in the enriched fecal specimen that interfered with amplification (34). When genomic DNA was purified from the recovered Salmonella by phenol-chloroform extraction and tested by PCR using both hisJ and invE-A primer sets, correctly sized DNA fragments were amplified. These results indicate that in some fecal specimens the method used to process the specimen is important, because if PCR inhibitors are not removed, a false-negative PCR result may be obtained.

Feces identified as containing Salmonella spp. by bacterial culture were more frequently identified as PCR positive when the hisJ primer set was used than when the invE-A primer set was used (sensitivities of 93.3 and 80%, respectively). However, PCR with the hisJ primers was more likely than PCR with the invE-A primers to produce an amplicon from feces that were negative by bacterial culture (specificities of 85.6 and 98.6%, respectively). The hisJ primer set appeared to be more sensitive than bacterial culture; use of these primers resulted in amplified products of the correct size (496 bp) in 16.9% of culture-negative samples. These results were initially considered false positives, and it was assumed that these PCR results were due to nonspecific amplification from a closely related bacterium with a DNA sequence that appeared to be identical in size to the authentic hisJ amplicon. Alternative explanations for these results are that amplification with hisJ primers detected levels of Salmonella spp. that were undetectable by the described culture methods or that the hisJ gene fragment was amplified from dead or damaged bacteria that were nonculturable. These two explanations are unlikely, since 37 (of 42) of the culture-negative specimens that were PCR positive with hisJ primers were PCR negative with the invE-A primers.

To determine whether the amplicons produced by PCR with the hisJ primers were truly false-positive reactions, i.e., they did not represent hisJ from Salmonella spp., the amplicons were subjected to DNA sequence and phylogenetic analyses. Six amplicons each from culture-positive and -negative samples were excised from agarose gels, purified using the QIAquick Gel Extraction kit (Qiagen, Inc., Valencia, Calif.), and submitted for DNA sequencing (DNAExpress, Colorado State University, Fort Collins, and Davis Sequencing, LLC, Davis, Calif.). The DNA sequences of 445 nucleotides from each of the six culture-negative and six culture-positive amplicons and S. enterica serovar Typhimurium hisJ (STHISJ locus, accession no. V01372.1 [16]) were aligned using CLUSTAL W (28), and consensus sequences corresponding to the culture-negative and culture-positive (designated negative consensus and positive consensus, respectively) specimens were generated. A nucleotide was included in the consensus sequence if it occurred in the same location in 50% of the amplicons from the clinical specimens. Examination of the aligned sequences revealed that it was possible to visually distinguish whether the amplicons were generated from either the culture-positive or the culture-negative specimens without prior knowledge of the culture results (data not shown). Amplicons from the culture-positive specimens were highly homologous to each other and to the reported DNA sequence of S. enterica serovar Typhimurium hisJ (98 to 99% identity). Amplicons from the culture-negative specimens shared a high degree of homology among themselves; however, there was a discernible difference (85 to 88% identity) from the reported hisJ sequence. The amplicons from culture-negative samples contained 69 nucleotide changes (over a length of 445 bp) compared to those from culture-positive specimens. Almost every nucleotide change occurred in the wobble region of the codon; however, none of these nucleotide changes caused truncation of the translated protein. Despite these nucleotide changes, there were only five amino acid substitutions in the hisJ amplicon from the six culture-negative samples. The possibility of an amino acid substitution resulting in a nonfunctional HisJ protein and inability to culture the bacterium is unlikely, since histidine transport systems (five total) are redundant in Salmonella spp. The high-affinity hisJQMP transport system uses either HisJ or ArgT as the periplasmic, substrate-binding protein that binds the membrane components for transport of l-histidine (4, 17, 35). In addition to these two high-affinity histidine transport systems (K_m of 10⁻⁸ and 10⁻⁶, respectively), Salmonella spp. have at least three additional low-affinity histidine transport systems. Therefore, even if HisJ were nonfunctional, the viability of Salmonella spp. should not be affected, since other histidine transport systems are available.

A phylogenetic analysis was performed to determine the relatedness of the hisJ-generated amplicons from the culture-positive and -negative samples. To do this, the DNA sequences of the culture-negative and -positive amplicons and the negative and positive consensus sequences were submitted to the National Center for Biotechnology Information (NCBI) (National Library of Medicine, National Institutes of Health, Bethesda, Md.) for a nucleotide BLAST (2, 3) search of the GenBank (NCBI), EMBL, DNA Databank of Japan (DDBJ), and Protein Data Bank (PDB) genetic sequence databases. S. enterica serovar Typhimurium hisJ sequences (STHISJ locus, accession no. V01372.1 [16]; STHISX locus, accession no. V01373.1 [17]; STYHISTO locus, accession no. J01805.1 [16]); Escherichia coli genomic DNA, Kohara clone 406, carrying the E. coli hisJ homolog (D90862 locus, accession no. D90862.1 [36]); E. coli K-12 MG1655 section 210 of 400 of the complete genome, carrying the E. coli hisJ homolog (AE000320 locus, accession no. AE000320.1 [6]); and E. coli hisJ (ECU47027 locus, accession no. U47027) (A. Joshi and G. F. Ames, unpublished data) were identified as having significant homology to both consensus sequences. An analysis by the unweighted pair group method using arithmetic averages (27) of the hisJ consensus sequences, the six culture-negative and six culture-positive hisJ sequences, and the six homologous nucleotide sequences obtained from the BLAST search identified E. coli K-12 MG1655 section 210 of 400 of the complete genome, carrying the E. coli hisJ homolog (AE000320 locus, accession no. AE000320.1 [6]); E. coli hisJ (ECU47027 locus, accession no. U47027; Joshi and Ames, unpublished data); and the E. coli hisJ homolog (D90862 locus, accession no. D90862.1 [36]) as outgroups for phylogenetic analysis. Distance, maximum-parsimony, and maximum-likelihood analyses were performed using the PAUP (phylogenetic analysis using parsimony) software of Swofford (version 4; Sinauer Associates, Sunderland, Mass.). All three analyses were off-strapped using 1,000 replicates. One representative DNA sequence from each of five groups (of those containing identical DNA sequences) was used for this analysis. The five groups were as follows: group 1, D90862 locus (E. coli genomic DNA, Kohara clone 406, carrying the E. coli hisJ homolog [36]) and AE000320 locus (E. coli hisJ homolog [6]); group 2, positive consensus sequence, 989023679, and 989486951; group 3, 98903933 and 98946992; group 4, STHISJ (16), STYHISTO (16), STHISX (S. enterica serovar Typhimurium hisJ [17]), and 98907916; group 5, 98902367625; and group 6, ECU47027 locus (E. coli hisJ [Joshi and Ames, unpublished data]). In addition, DNA sequences of the culture-negative amplicons (from specimens 978323182, 97832372, 978326281, 978326282, 989006403, and 989024783) were included in the analysis, since they differed from each other by 1 to 2 nucleotides. Duplicate identical DNA sequences were not included in these phylogenetic analyses.

The neighbor-joining dendrogram generated from these data revealed two well-supported, distinct clades of bacteria—one clade corresponding to the six culture-positive specimens and a second clade composed of the six culture-negative (PCR-positive) specimens (Fig. 2). This analysis indicated by strong bootstrap support (100%) that the two clades were more closely related to each other (and thus to S. enterica serovar Typhimurium hisJ) than to the E. coli hisJ genes in the NCBI databases. One clade represents culturable Salmonella spp., and the second clade may correspond to a PCR-detectable bacterium that was not identified by our culture methods and is highly related (or perhaps identical) to Salmonella spp. Failure to culture these Salmonella-like bacteria from these fecal specimens may be due to the fact that they are damaged or have special nutritional requirements that are not provided by the media used for enrichment and selection. Alternatively, the bacterial DNA amplified by the hisJ primers may originate from a bacterium that is closely related to Salmonella spp. and whose hisJ sequence is not currently available in the gene and protein databases. PCR using primers for additional Salmonella-specific genes and genetic analysis of the resulting amplicons need to be developed before conclusions can be made on the identity of the bacterium contributing to the amplification of the hisJ homolog. Although analysis of this study is limited by information available in the gene and protein databases, we propose that culture-negative specimens contain a bacterium that is closely related or identical to Salmonella spp.

FIG. 2.

Relationship of hisJ genes from Salmonella culture-positive and culture-negative enriched fecal specimens obtained by phylogenetic analysis of hisJ sequence data. The numbers at each branch point indicate the percentage of bootstrap replications. E. coli hisJ homolog (AE000320 locus, accession no. AE000320.1 [6]) and E. coli hisJ (ECU47027 locus, accession no. U47027 [Joshi and Ames, unpublished data]) were designated as outgroups.

In conclusion, in our laboratory, PCR of feces with the S. enterica serovar Typhimurium hisJ or invE-A primers resulted in detection of a higher or lower number, respectively, of Salmonella-positive specimens than did conventional bacterial culture. At this time, PCR using invE-A primers could be used as a preliminary screening procedure to rapidly identify animals that are shedding large numbers of Salmonella spp., with the realization that false negatives may occur when compared to culture results.

In addition, during the course of this study we determined that the hisJ primer set detected a bacterium that represents either a subset of Salmonella spp. or a related, Salmonella-like bacterium, previously undescribed in feces. Therefore, PCR with the hisJ primer set for identification of Salmonella spp. in fecal samples should be used with caution, since the identity of the bacterium that is being amplified with the hisJ primers is unclear at this time. Future studies are necessary to identify the bacterium that is being amplified by PCR with the hisJ primers, since these primers may be amplifying DNA from an avirulent Salmonella sp. or a damaged, altered form of Salmonella spp.